Re: [PATCH 5/5] xfs: reinitialize rm_flags when unpacking an offset into an rmap irec

["Darrick J. Wong" <darrick.wong@oracle.com>]

On Thu, Aug 29, 2019 at 12:29:57AM -0700, Christoph Hellwig wrote:
> On Mon, Aug 26, 2019 at 02:49:46PM -0700, Darrick J. Wong wrote:
> > From: Darrick J. Wong <darrick.wong@oracle.com>
> > 
> > In xfs_rmap_irec_offset_unpack, we should always clear the contents of
> > rm_flags before we begin unpacking the encoded (ondisk) offset into the
> > incore rm_offset and incore rm_flags fields.  Remove the open-coded
> > field zeroing as this encourages api misuse.
> 
> This one doesn't fit the series' theme, does it? :)

Nope, there's always one cling-on patch. :/

> > +++ b/fs/xfs/libxfs/xfs_rmap.c
> > @@ -168,7 +168,6 @@ xfs_rmap_btrec_to_irec(
> >  	union xfs_btree_rec	*rec,
> >  	struct xfs_rmap_irec	*irec)
> >  {
> > -	irec->rm_flags = 0;
> >  	irec->rm_startblock = be32_to_cpu(rec->rmap.rm_startblock);
> >  	irec->rm_blockcount = be32_to_cpu(rec->rmap.rm_blockcount);
> >  	irec->rm_owner = be64_to_cpu(rec->rmap.rm_owner);
> > diff --git a/fs/xfs/libxfs/xfs_rmap.h b/fs/xfs/libxfs/xfs_rmap.h
> > index 0c2c3cb73429..abe633403fd1 100644
> > --- a/fs/xfs/libxfs/xfs_rmap.h
> > +++ b/fs/xfs/libxfs/xfs_rmap.h
> > @@ -68,6 +68,7 @@ xfs_rmap_irec_offset_unpack(
> >  	if (offset & ~(XFS_RMAP_OFF_MASK | XFS_RMAP_OFF_FLAGS))
> >  		return -EFSCORRUPTED;
> >  	irec->rm_offset = XFS_RMAP_OFF(offset);
> > +	irec->rm_flags = 0;
> 
> The change looks sensible-ish.  But why do we even have a separate
> xfs_rmap_irec_offset_unpack with a single caller nd out of the
> way in a header?  Wouldn't it make sense to just merge the two
> functions?

xfs_repair uses libxfs_rmap_irec_offset_unpack, which is why it's a
separate function.

--D